Session Leader: Jens Østergaard

Notetaker: Gabrielle Benefield

Organizational Patterns use a variety of data-gathering techniques that support retrospectives. The techniques have roots in (non-pathological) dissociative psychology and social network theory. A long-running Bell Labs research program gleaned results from over 100 retrospectives to build a model of successful contemporary software development practices. The results were the foundations both of Scrum sprints and of the structures of XP.

An organizational pattern exercise is overseen by a facilitator. The intervention is broken down into two main activities: role-play, and a feedback session. Both of these have retrospective elements. In addition, supporting one-on-one interviews, personality analyses, and other techniques might supplement the intervention.

“Retrospective” means “looking back,” and there are many ways to reflect on one's history. Monadic techniques employ individuals reflecting on their own experience, usually through a diary or a single-person reflection. Diadic techniques employ an external facilitator or interviewer who works with individuals or small groups. In triadic techniques, the facilitator observes and records interactions between the subjects, with only light facilitation or intervention. The advantage of triadic techniques is that a large group helps keep individuals within the context of a shared perspective; smaller groups can remember things differently, which makes integration difficult.

The technique takes the form of a role-play. The role-play revisits one or more past episodes as individual time lines. Individuals fill out CRC cards in a role-play, speaking only about their own role when that role is engaged in the time line.

After the role-plays, we ask subjects to weight the strength of the collaborations: high, medium, or low. We feed those data into a simulated annealing algorithm that produces an “organizational chart” of the subjects, showing communication adjacency and clique structure:

This picture is called a sociogram, and was pioneered as part of the science of social network theory by Moreno in the early 20th century. As Andreas noted, this structure is analogous to an “organizational landscape” exercise.

We then analyze the results and write a report that includes notes on the interactions, the sociogram, and other sociometric analyses. We present those results to the team in a second meeting, a feedback session. We call this a “feedback session” because we give a short oral version of the report. However, the goal of the meeting is that the subjects reflect on the results. The feedback is not cut-and-dried, but is open-ended and is designed to encourage reflection. This meeting is more like a “traditional” retrospective session in that it involves reflection. The reflection is supported by the fact that the team is distanced from the events themselves and has been removed from remembering the time lines and feelings; they are focusing instead on the sociogram models. These models help the group take a different, often objective look at themselves in a way that is separate from the role-play.

Organizational learning takes place both in the role-play and in the feedback session.

• Papers on Organizational Patterns can be found on Google.
• The Scrum community (Gertrud Bjørnvig, Dina Friis, and Jens Østergaard, supported by Jim Coplien and Jeff Sutherland) are currently doing research to show that organizational pattern conformance is a good litmus test for Scrum practices.